From mboxrd@z Thu Jan 1 00:00:00 1970 From: Guenter Roeck Date: Mon, 11 May 2015 13:12:27 +0000 Subject: Re: [lm-sensors] Additional PWM driver support for w83792d Message-Id: <5550AABB.2030006@roeck-us.net> List-Id: References: <003901d08a57$bc46f8a0$34d4e9e0$@hiddenengine.co.uk> In-Reply-To: <003901d08a57$bc46f8a0$34d4e9e0$@hiddenengine.co.uk> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: lm-sensors@vger.kernel.org On 05/11/2015 03:45 AM, Jean Delvare wrote: > On Sun, 10 May 2015 09:01:25 -0700, Guenter Roeck wrote: >> On 05/10/2015 06:42 AM, vt8231@hiddenengine.co.uk wrote: >> [ ... ] >>>>> >>>>> Last note: registers 0xA3-0xA6 have extra configuration bits "Sync >>>>> T1/2/3". Maybe the driver should handle them but I am not sure how. It >>>>> could be that the extra outputs should only be exposed to user-space if >>>>> these bits are 0 (stand alone.) Guenter, any idea/opinion on this? >>>> >>>> How about using pwm[4567]_enable ? If I understand correctly, the possible >>>> modes would be manual or sync(x). In this case we could have 1 (manual), >>>> 2 (sync with fan1), 3 (sync with fan2), and 4 (sync with fan3), with the >>>> caveat that the sync settings only make sense if the matching pwmX_enable >>>> is set to thermal cruise mode. >>>> >>>> Does this make sense ? > > Not really. The problem is that for now I'm not sure what "sync" really > refers to. The datasheet mentions temperature channels, and as far as I > can see each temperature channel is hard-bound to a specific fan > output. So I suspect that what these configuration bit really mean is > that the pwm4 output (for example) mirrors the pwm1 output, i.e. pwm4 > has no independent existence. In which case it's better to not expose > it at all. > > If the BIOS specifically configured these bits, there must be a reason > and that reason would be the way the fans are connected to the chipset. > Better not change it. > > If we really want to expose these bits then abusing pwmX_enable the way > you suggested is still not correct, pwmX_auto_channels_temp would be > better suited. But then again I don't think it adds any value if the > extra PWM outputs only mirror already existing PWM outputs. > Ok, fine with me. Guenter _______________________________________________ lm-sensors mailing list lm-sensors@lm-sensors.org http://lists.lm-sensors.org/mailman/listinfo/lm-sensors